Deepstream 5.0, were changes made to the batch size settings behavior?

[prominence_ai]

Please provide complete information as applicable to your setup.

• Hardware Platform (Jetson / GPU)
Jetson
• DeepStream Version
5.0
• JetPack Version (valid for Jetson only)
4.4
• TensorRT Version
7.1

Did the behavior/rules change in the way the batch size values are used? … it seems there is an unfamiliar messages in the console about “because the batch batch_size was explicitly set”, that I don’t recall seeing before?

Do we still need to set the batch_size for the streammux, gie’s, etc. … which has always been a little problematic when dynamically adding and removing sources/sinks ?? I couldn’t find anything in the release notes.

Thanks,
Robert.

[bcao]

@prominence_ai

Did the behavior/rules change in the way the batch size values are used?

No.

@prominence_ai

Do we still need to set the batch_size for the streammux, gie’s, etc

Yes, still needed.

In addition, If mux batch size > num sources, mux might wait till timeout before pushing the batch. This can be fixed by using timeout slightly less than 1/(fastest fps)

[prominence_ai]

@bcao

This can be fixed by using timeout slightly less than 1/(fastest fps)

Thanks @baco, for clarifying the above. Very much appreciated!